Fault Diagnosis and Prognosis Based on Deep Learning and Transfer Learning
摘要
The effective operation of fault diagnosis and prognosis algorithms relies upon effective fault or performance degradation feature extraction and a large amount of model training data. However, the engineering applications of fault diagnosis and prognosis technology face great difficulties and challenges due to the high uncertainty and costs of artificial feature engineering, failure of diagnosis and prognosis models for a lack of fault data and high costs of fault data acquisition.